What MMM Does
Marketing mix modelling looks at your total sales over time alongside spend on each channel, plus things like seasonality, pricing and promotions. It then estimates how much each channel contributed. Because it uses aggregate numbers, not individual tracking, privacy changes and blocked cookies don't affect it.
It's built for strategic questions: how should next year's budget split between search, social, TV and email?
MMM vs Multi-Touch Attribution
| Marketing mix modelling | Multi-touch attribution | |
|---|---|---|
| Data | Weekly or monthly spend and sales totals | Clicks and conversions per person |
| Covers offline channels | Yes | Mostly no |
| Detail level | Channels | Campaigns, ads, keywords |
| Speed | Slow; refreshed periodically | Near real time |
| Affected by privacy limits | Little | Yes |
| Best for | Annual and quarterly budget splits | Daily optimisation |
They answer different questions, so mature teams use both, and calibrate MMM with incrementality tests.
What MMM Needs
- History. Typically a couple of years of weekly data.
- Variation. If spend never changes, the model can't see its effect.
- Clean inputs. Spend by channel, sales, prices, promotions and major events.
- Skills. Someone who understands the statistics, or a vendor who does.
Free and Paid Options
Google Meridian
Google's open-source MMM framework.
Meta Robyn
Meta's open-source MMM package, released in 2020.
PyMC-Marketing
An open-source Bayesian toolkit in Python.
Vendor Add-Ons
Some attribution platforms, such as Northbeam on higher plans, offer MMM as an add-on.
Open-source removes the licence fee, not the work. Expect to spend real time preparing data and checking results.
When MMM Makes Sense
- You spend across many channels, including offline ones like TV, radio or print.
- Your budget is large enough that a better channel split is worth serious effort.
- You have enough history and spend variation to model.
For most small and mid-sized advertisers, we'd start with accurate attribution. It answers the daily questions about which ads to scale or cut, and it's where most budget is won or lost. See multi-touch attribution and the best attribution software.
Frequently Asked Questions
What is marketing mix modeling?
Marketing mix modeling (MMM) is a statistical method that estimates how much each marketing channel contributes to sales, using weekly or monthly spend and revenue data rather than tracking individual people.
How is MMM different from multi-touch attribution?
MMM works on aggregate data across all channels, including offline ones, and helps split budgets. Multi-touch attribution follows individual journeys and helps compare campaigns, ads and keywords day to day.
Are there free MMM tools?
Yes. Google's Meridian, Meta's Robyn and PyMC-Marketing are open-source frameworks. They still need clean data and analytical skills to run well.
How much data does MMM need?
Typically a couple of years of weekly spend and sales data with enough variation in spend to see effects. New or small businesses often don't have enough yet.
Should a small business use MMM?
Usually not first. Accurate attribution and occasional incrementality tests deliver more value for smaller advertisers. MMM becomes useful once you spend across many channels, including offline.